15:13:46stassatsfor some reason code-header-words is two-word aligned on not x86-64
15:18:12stassatsok, code-boxed-words-align is different, so that's expected and only a test needs to be adjusted
15:18:40stassats(+ 2 #+(or x86 x86-64) -1) is a weird way to word it though
15:23:35stassatsthe sigfpe problem -- the ppc64 kernel still being broken, even though my previous C test case no longer fails
15:23:37ShinmeraSo on windows allocating foreign memory, dumping an executable, and then reading that memory back gives very bad results. (typically memory faults) Any advice aside from "don't do that"? The memory in question is not in a foreign library, just cffi:foreign-alloc.
16:38:16scymtymfor me, the elfcore test hits primitive halt possibly because shrinkwrap-sbcl seems to ignore toplevel options
17:15:29scymtymKrystof: i have been annoyed with the cost of applicable method checking in (safe) CALL-NEXT-METHOD calls (example: https://techfak.de/~jmoringe/clim.flamegraph-medium-draw-rectangle-star.png ). i came up with this slightly crazy way of doing the checks efficiently: https://github.com/scymtym/sbcl/commit/c0997d77f0b198a3b70cdbae1c68514681eb491a . do you think this could work?